Policy entrepreneurship in UK central government: The behavioural insights team and the use of randomized controlled trials

Introduction

Journalists often pronounce the Behavioural Insights Team (BIT) as a success. ¹ This is a surprisingly positive press for the work of a small body of civil servants and advisers, based in the Cabinet Office, who have very little money and no powers to do its job. In fact, it might be thought that the use of psychological research to encourage citizens to behave in better ways would attract the ire of the populist press, resistant as it is to paternalist measures coming from the state. Not only has the team won the support of the media, it has pioneered the application of behavioural insights in government. As this article will show, it has helped introduce a large number of initiatives and successfully promoted the use of randomized controlled trials (RCTs).

This success has been achieved in an unpropitious context. At the top of the civil service, it is usually hard to promote new ideas outside the main corridors of power as past advisers to prime ministers have found to their cost. Consider those who ran the ill-fated Central Policy Review Staff (CPRS) (Blackstone and Plowden, 1990); or Bernard Donoughue, who had to fight endless internal battles to set up and operate the Prime Minister’s Policy Unit during the 1974–1979 Labour government (Donoughue, 2005); or the advisers to Tony Blair, who could not function effectively in such a fast-changing and fractious environment (Rawnsley, 2000, 2010). While prime ministers often desire to implement radical policy changes, in practice such initiatives are usually short-lived, and political leaders get ground down by the pace of government, the power of the insiders, the constant need for good headlines, the endless internal battles and the power of producer groups— what Tony Blair called the ‘scars on my back’. ² Moreover, the bureaucracy at the centre of government is often resistant to change and is ready to oppose new units that might drive innovation (Kelman, 2005). In this context, the success of BIT is intriguing and suggests there are some special factors at work, which could throw some light on more general conditions for innovation in public sector organizations, especially those at the centre of government.

The aim of this paper is to examine the history and operation of the team to uncover what factors have caused its success, applying insights from organizational theory and studies of private sector management. It adapts the framework of change agents to identify the likely factors at work (Kanter, 1983), finding that not all of them need to be present. By focusing on necessary rather than sufficient conditions, this account of the team’s work can help produce a more parsimonious account of innovation adoption in central government.

The paper takes the following structure. The first part gives a history of the team and outlines its successes. The second section is a review of academic approaches to innovation in the private and public sectors. The third applies the change agent framework to the activities of the team. The final part offers some general conclusions of what can be learnt from the application of the framework.

The behavioural insights team

The UK coalition government elected in 2010 set up the Behavioural Insights Team, sometimes referred to as ‘the nudge unit’ or just BIT. The unit ‘draws on insights from academic research in behavioural economics and psychology, to apply them to public policy making’. ³ It aims to find ways to encourage citizens to adopt more pro-social behaviours and to make better choices. It was created in June 2010 as a unit within the Cabinet Office. It comprised at first seven officials; now it has expanded to 13. It takes advice from experts, such as Richard Thaler, and set up an academic advisory panel.

Early in its life, the team was influential in persuading the Driver and Vehicle Licensing Agency (DVLA) to require those who are renewing their driving licence to choose whether to agree that their organs may be donated in the event of their death. The team has pioneered a number of other reforms, which have appeared in papers, such as on energy use (Cabinet Office et al., 2011), which reports work with the private sector to try out different kinds of incentives for consumers to change their behaviour. BIT worked with the Department for Business Innovation and Skills (BIS) on a consumer empowerment strategy, Better Choices: Better Deals (April 2011), and also with the Department of Energy and Climate Change (DECC) on energy saving, aiding in the redesign of Energy Performance Certificates (EPCs). The team has done research into energy insulation, finding that when people were offered loft clearance schemes at cost price, there was a fivefold increase in loft insulation. The main obstacle to people getting loft insulation is not cost but the difficulty of sorting out their lofts. The unit produced a paper on health, which reports the work on smoking cessation, and then a paper on charitable giving, jointly written with the Office for Civil Society in the Cabinet (Cabinet Office, 2010). The team has worked on measures to encourage people back into employment working with Job Centres to improve advice. There is also a stream of work on charitable giving, working with financial institutions to to encourage employees to donate portions of salary (Cabinet Office, 2013).

One of the key activities of the unit is its use of RCTs to test out interventions, which has become more a feature of its work as the team has settled in and developed its approach. To this end, the team published in 2012 Test, Learn, Adapt: Developing Public Policy with Randomised Controlled Trials. Written with academics, this is a guide about how to do experiments (Haynes et al., 2012) and was launched at a conference in June 2012.

The team worked with Her Majesty’s Revenue & Customs (HMRC) in February 2011 to pioneer different wordings for the reminder of tax returns, which used RCTs. HMRC has now carried out a set of trials on tax reminders coordinated by Michael Hallsworth. BIT worked with the HM Courts & Tribunals Service to increase the number of fine payments through use of personalized text message prompts, reducing the number of bailiff interventions by 150,000. The team has done work with DVLA to find out whether a photo of the owner’s car would encourage people who are behind with registration to update. Many of these findings are summarized in Fraud, error and debt: behavioural insights team paper (Cabinet Office, 2012). The team’s documents claim that the reforms of individual decision-making has identified public savings, such as from court fines and tax reminders, at least £300m over a five-year period. This is done by using the effect sizes measuring the impact of the intervention from the RCTs and then extrapolating them to the volume of transactions in each activity, such as the benefits of early payment of taxation (Behavioural Insights Team, 2012).

The team operates collaboratively which largely reflects its small size and the need for willing partners to carry out its interventions. The team works beyond central government departments and deals directly with local authorities, but it would also be fair to say the most of the team’s work is with central government departments because they have day-to-day dealings with the Cabinet Office and its ministers, and where central government has the power and legitimacy to act.

The team has been met with enthusiasm across Whitehall, partly as a result of the briefings and events the team holds. At the same time, there is a considerable amount of contact between team members and the rest of the world, such as academics making contact, resulting in stream of eminent visitors to 10 Downing Street, and many by bureaucrats from other countries. BIT works with other governments, such as Government of New South Wales, Australia, where a member is seconded in 2012–2013. The team has a project with Freebridge Housing Association, which is seeking to use behavioural insights for the redevelopment of a housing estate in King’s Lynn. Another feature of the team’s work is the positive press it has received, which got much better during the life of the team. This was not the case at first, such as when it was branded as Cameron’s vanity project, and Brendon O’Neill called for the illiberal nudge industry to ‘push off’. ⁴ But by 2011 it had got praise in newspapers right across the political spectrum. ⁵

Of course, no unit in the febrile environment of central government is going to operate without controversy and has to work in the context of competition between departments and ministers, where they are alternative sources of advice and considerable pressures to claim credit. So the work on texting with mobile phones was done with the courts service, but the Ministry of Justice and its research team were less involved. Parts of HMRC are very supportive of the trials that can be done, but other parts of this vast bureaucracy are much less interested in this agenda. The unit did not manage (at first) to persuade DVLA to the presentation of the option of organ donations by a randomized trial in spite of trying. Not all the experiments worked or produced the expected results, such as the trial of signature placement with a local authority that did not confirm Dan Arieley’s argument that signatures at the top of documents increased compliance. The unit even got some bad press for its job centre experiments. ⁶

Nonetheless, when considered alongside the other kinds of initiative across government, the conclusion to draw is that there is an increasing amount of activity at the centre to encourage smarter behaviour change policies, much of it encouraged by the work of the team. BIT is developing its approach and has become more experienced at promoting behaviour change, especially from its use of robust evidence from RCTs. The unit was set up with sunset clause that provided for its closure in 2012. Showing its importance for government, the team has successfully completed its sunset review. It now continues into the future with a broader remit and has the ability to work overseas with other partners. In December 2012 the team held an event to announce that is was going to operate as a mutual body at arms length from government. Such an extension is a mark of the team’s success and of the continuing interest of the government in research and policies on behaviour change. The government will hold a half share in the scheme and still continue to commission work, but BIT can expand its activities. The unit claims this is because of extensive demand for its services that is not able to meet all these in its current organisational form.

Interestingly, the unit has fared much better than models in other countries. President Obama appointed Cass Sunstein, one of the authors of Nudge (Thaler and Sunstein 2008), to head up the Office of Information and Regulatory Affairs, but he has since stood down. In France, Centre for Strategic Analysis of the Prime Minister employed a behaviour science expert, Olivier Oullier, as an adviser. But under President Hollande this no longer happens. In fact, what happened in 2013 is the transfer of the BIT model to other jurisdictions, such a new unit in the White House. ⁷ So there must be a secret ingredient to BIT’s success.

Innovation

Work in organizational theory examines the conditions that can help organizations become more innovative. In the private sector, such innovation is associated with new products and higher profitability (Argyris, 1993; Nooteboom, 2000), but in the public sector is often about policy changes and better approaches to public sector management (Kelman, 2005; NAO, 2009). Writers in this literature outline the forces for conservatism within organizations, which are due to the power of routines, psychological factors and standard operating procedures, which tend to benefit those in power. Existing power holders may resist new policies because they may be associated with younger post holders whose careers might benefit. To this can be arrayed forces for innovation that can overcome such resistance, whose advocates are often concentrated in small groups in the organization’s bureaucracy. But these reform groups need nurturing and must build a successful coalition to overcome change. In addition, there is long-held assumption that innovation is hard to achieve in the public sector because of the lack of the profit motive. Moreover, lines of accountability to political office holders mean that bureaucrats have limited discretion to innovate independently. Nonetheless, in the right conditions innovation can occur (Borins, 2002), and it is possible to read across the private and public sectors, taking note of the context of each when assessing what factors drive innovation.

Moss Kanter in The Change Master (1983) identifies special conditions whereby what are called change agents can become more prominent in an organization. Change agents or masters can pioneer innovations, which can then diffuse within the organization. Kanter calls innovations new streams that need to be managed. Key is the ability of the leaders of the organization to allow enough delegation so these units can operate: ‘The degree to which the opportunity to use power effectively is granted to or withheld from individuals is one operative difference between those companies which stagnate and those which innovate’ (89).

These units have several names, but the one that has become celebrated in the literature is skunkworks. The term was reported by Rogers in the Diffusion of Innovations (Rogers, 1995: 139–140) to describe a section in an organization charged with coming up with an innovation, and also popularized by Peters and Waterman (Peters, 1982) in their famous management book, In Search of Excellence as well as appearing in the academic literature on organizational innovation (Bommer et al., 2009; Fosfuri and Thomas, 2009). The derivation, Skonk Works, featured in the Al Capp comic strip Li’l Abner, which was popular in the 1940s. Skonk Works—later Skunk Works—become the nickname for the Advanced Development Projects Division in Burbank, California, which designed Lockkeed’s P-80 Shooting Star during the Second World War (Jenkins, 2001). ⁸ The unit was housed out in a windowless block next to a plastics factory, where its workers were given a large amount of autonomy by the senior management who just expected them to just get on with the job and to come up with the innovation. Over time the Skunk Works Unit was charged with other product innovations. Then other technology companies copied the idea. For example, Steve Jobs of Apple created a Texaco Towers team in this way. Such units are often celebrated in the media. For example, in the film The Dam Busters civil servants allowed the eccentric inventor Barnes Wallis to form a group of inventors and aviators to design and test the bouncing bomb.

In these accounts, it is possible to identify several conditions for the effective success of such innovations, which can either be a feature of the organization or come from the management style that a chief executive introduces. The units are managed with:

less hierarchy;

Kanter (1989) talks about the need to manage the tensions between the mainstreams and new streams, each of which require different sorts of management. The key difficulty is the transfer of the new stream back to the mainstream, what can be called mainstreaming, whereby the innovative practice becomes part of the normal running of the organisation; but in practice this is very hard to achieve because of the tendency of old practices and routines to reassert themselves, and for innovations to get watered down, which often involves re-labeling old practices as new ones.

There is another aspect of Kanter’s framework that deserves particular attention: the interaction between these conditions and the personal characteristics of the managers who are charged with sponsoring or carrying out innovations. Kanter pays great attention to this in her later works which are more about how individuals might learn various techniques to become change agents (Kanter, 1999), but what is important is whether the structures she discusses interact with innovators and risk-takers. Put simply, if the right people are placed in the appropriate structures then innovation can take place.

There are a number of difficulties with this kind of work in that it relies on selecting successful companies and then making an inference from the incidence of these factors, a classic example of selection bias. The reader is not informed whether failing companies also adapted the same measures. The skunkworks examples are also very selective, and it very hard to work out what is rhetorical from what is true: these are the stories that companies like to tell. It is very hard to make a causal inference as there are so many factors at work, and it is not clear from reading these academic papers which element is predominant. One advantage of a case study is that it can examine what is the relative incidence of various factors. With BIT there is a clear history of innovation, so it is possible in one case to see what conditions apply and what do not. Of course, it is not possible to confirm or falsify elements of the framework, but it is possible to use it to gain further understanding of what causes innovation.

Explaining the success of BIT

In this section, theory and practice come together in a review of the experience of BIT, as illuminated by the factors listed above. The first is whether there is less hierarchy with the team. Here there are elements of the team that fit into the conventional hierarchy of the civil service, as the fact that most of the staffers were seconded from other units, such as the deputy director of the team, Owain Service. It operates under the management procedures of the Cabinet Office, being subject to the direction of a steering committee, chaired by the Cabinet Secretary. Normal rules apply for appointments and its general operation. There is nothing special in its procedures and it compares to other units of the similar kind, such as the Office for Civil Society. But it is clear that the unit has a relatively free hand to define its work and the steering was light touch. David Halpern, who had previously worked for Labour in the Prime Minister’s Strategy Unit, likes to work across institutional structures, prefers to deal closely with the politicians, and is adept at summarizing research findings in an attractive and common sense way that is very appealing to a practitioner audience.

Though there is a director and deputy director, and line management operates, the team is largely flat in hierarchy and its members like work closely together. The layout of the Cabinet Office encourages this with its long lines of shared desks. Halpern works between the Cabinet Office and a base in Number 10. The team employs outsiders, such as Laura Haynes, who has an academic background, and engaged a series of short-term research fellows, who are often current PhD students, helped with support from the Economic and Social Research Council. The team works with whoever is willing to do projects with it, usually in a collaborative rather than hierarchical manner. To design and analyze the trials, it sometimes draws on the expertise of academics on its advisory panel. The mutual organization may support the non-hierarchical approach of the team as it can work outside the normal procedures governing the civil service.

The second factor is linked to the first, which is a different framework of performance evaluation. There was no strict framework in place: BIT was not set targets and was left to formulate its own initiatives subject to the approval of the steering board. The politicians were even prepared to let the unit fail, a view which shocked the press. ¹⁰

The third factor of the role of senior champions is particularly important in central government. Here the role of the Prime Minster is pertinent, given his interest in behaviour change and wish to foster the project, The Big Society. As important were other champions: one was the Prime Minister’s former director of strategy, Steve Hilton, who is interested in behaviour change and innovation in central government; the other is the former Cabinet Secretary, Gus O’Donnell, who was an academic economist, is enthused by the behavioural economics agenda and takes a personal interest in the work of the team. These are powerful people who are able to open doors and offer protection. What is interesting from the change agent perspective is that both left government in 2012, but without affecting the team’s work. Steve Hilton left in May 2012 and O’Donnell retired in December 2012. But the unit continued effectively in 2013. It enjoys good political support from Cabinet Office ministers, Francis Maude and Oliver Letwin.

The fourth factor—being funded differently—does not apply to the team’s work as it works within the civil service. There is no different system of cost control. The move to more private working may increase financial flexibility. Fifth, there is no separate space for the team as it occupies space amid the anonymous ranks of desks at One Whitehall with just a small poster to indicate its presence. There is no special building redolent of the Skunk Works Unit. Given the political environment it is important for the unit to be plugged into the decision-making machinery and to use political access effectively. A separate location (as CPRS had) might not have worked in these circumstances. Even though the unit has freedom, it probably works to shorter timescales than innovation management in the private sector, partly for political reasons even though policy has been more stable under the Coalition with the five-year fixed-term parliament and the Coalition Agreement, which has encouraged policy-makers to focus on implementation and improving policy.

Sixth, the time cycle is not relevant either since there was a two-year sunset clause on the unit’s work. The team had a slow start, and experienced some changes in members, such as the departure of Paul Dolan in 2011. In the second year of the team’s work it made its key successes and carried out the RCTs that are the basis of its current reputation.

Seventh, low staff turnover has been a feature of the team’s work and the team has remained stable since 2011 with the same members in place, which is one of the conditions for such units. This is in contrast to much of the Cabinet Office and other parts of government, which have experienced high staff turnover in recent years. The reasons for this are probably because members of the team like to work together, and there is a strong espirit de corps in the unit, which is in turn sustained by its relative longevity.

Discussion and conclusion

BIT has all the hallmarks of a change agent: it works in an alternative stream from other units of government, it promotes new ideas and ways of working, and it has been successful in promoting innovation. The literature on organizational change provides a good description of how the unit was set up and operated, how senior sponsorship allowed it to do new things and operate in part against the grain. One assumes that the politicians and civil servants knew what they were doing when they hired Halpern to direct the unit as he had performed a similar role with the Labour government. The Cabinet Secretary must have expected that the unit would attempt to drive innovation, and this fitted with his outlook as an unconventional civil servant. In terms of the conditions for a successful change agent, many of them are in place, such as working with less hierarchy, the relative freedom from conventional procedures, and the existence of a stable and small group of employees. More interesting are the factors that do not appear to be so important, such as not having a separate building, and the constraints on the time cycle. Also, with the departure of O’Donnell and Hilton, the need for senior protection seems less important than the literature suggests, though of course the senior protector supreme in the form of the Prime Minister is still in place. It may be the case that the lack of success of other Coalition government policies is partly responsible for this protection. Halpern and his unit are able to deliver to the Prime Minister successes in an otherwise bleak environment for the government. The costs of operating the unit are low, and there are few risks, especially since journalists like it too. One can imagine why the politicians are happy to sponsor it and that other parts of government are content to follow this central lead.

So the conditions identified by the change agent literature seem only partially applicable in this case, which is not surprising given the special and highly political content of UK central government. But there is enough that is recognizable from the change agent framework to suggest that it is something analogous at work here, in particular the idea of giving a small team the autonomy and protection to designs and promote innovations. The team model needs careful preparation and in particular requires the right balance of skills and people to do the job effectively. The team has to make friends and build support. But it shows what can be done with a modest level of investment by the centre. Whether such units are time limited is beyond the scope of this article, and it is clear that bureaucratic routines and demands of governing take priority in the long term. But when there is the right balance of environment, structure and people it is possible to produce more innovation at the centre of British government.